Thank you very much, Mark! Maybe all that needs to be done is a more
up-to-date wiki on including WaveLab in Octave? Of course, if someone
clones Matlab's wavelet toolbox, that may be optional... Did you do
this is in the Windows version of Octave (3.0.1)? I currently use a
very powerful software called IDL (Interactive Data Language) but it's
missing several important wavelet capabilities, such as wavelet packets
and matching pursuit. What's kept me from moving away from it earlier
is it has 64-bit addressing, suitable for the huge datasets I have to
analyze. For this reason, too, I've been looking at Scilab, but I
think WaveLab is more capable than Scilab's SWT. I really like
qt-Octave.
Mostly I study turbulence, compression, and denoising (places where
wavelets have grown significantly in the last decade). I will try your
script as soon as I can, and I think I'll try contacting the WaveLab
folks, researcher-to-researcher. Sometimes that helps a lot.
The Best,
PattiMichelle
Marc Normandin wrote:
code, or the reputation of its authors. I think all of these points
have significant merit: wavelets have garnered significant attention
across many disciples, WaveLab was developed and is still used despite
the existence of Matlab's wavelet toolbox, and the WaveLab developers
are prominent and highly respected in their field.
However, one must keep in mind that these are not sufficient reasons for
inclusion in Octave. Wavelets may be used often and by many, but
WaveLab is a large body of code to take on for functionality that many
(probably most) Octave users won't use. IMHO, it would be better suited
as an add-on package from Octave-Forge.
Licensing is a separate but perhaps more important issue. WaveLab is
released as "freeware", but I couldn't to find any license for the
release. The items I could find appear to have some elements of concern:
To paraphrase Wavelab850/Documentation/COPYING.m, redistribution is
permitted but the original work must be included in its entirety and the
original authors retain copyright on all redistributions, whether
verbatim or modified works.
And from the Introduction section of documentation available at
http://www-stat.stanford.edu/~wavelab/Wavelab_850/AboutWaveLab.pdf,
"The library is available free of charge over the Internet by WWW
access; instructions are given below. The material is, however,
copyrighted, so that advance permission is required for
any commercial use."
I am not a lawyer, but these terms seem to conflict with Octave's
license, the GNU GPL. Perhaps a "non-free" Octave-Forge package (like
Arpack and spline-gcvspl) would be be a possibility.
I don't dispute the importance of wavelets, the quality of the WaveLab
I've been trying to muck through WaveLabOnOctave, but I'm such an Octave
noob - sometimes an Octave guru can do in 5 minutes what would take a
noob a month (or more) to understand and implement.
I tinkered a bit with WavePath.m, the file that adds the WaveLab code to
your [Octave] path. I made two changes: (1) eliminated the dependence
on the "matlabroot" variable by simply requiring that you run the script
from the WaveLab directory, and (2) removed some strange markup that
existed in the original file. The modified WavePath.m is attached.
Executing this file seems to successfully add all of the WaveLab stuff
to the Octave path on my machine. I didn't try to compile the mex
files, but the documentation indicates that the functions should still
work, albeit more slowly. An attempt to run one of the demos failed, it
seems that the demo is graphically based and incompatible with Octave
graphics (at least with the gnuplot backend that I'm using). So the
demos may be unavailable to you, but the functions that actually perform
various analyses should be accessible.
Does this help any?
|